SOULMASK · SERVER REQUIREMENTS · NOVEMBER 2026

Soulmask Server Requirements: RAM, CPU and Player Slots

The official line is 16GB of RAM and 2-4 CPU cores per map, but what decides whether your Soulmask server stays stable is the world: recruited tribesmen, base size, world age and how long it runs between restarts. Here is how each one uses RAM and CPU, how much to buy for your group, and how to tell which resource is running out.

RAM

12GB can work for a small new world, but it is tight. 16GB gives proper headroom for world growth, save spikes and the known memory creep between restarts.

CPU

The official guidance is 2-4 cores per process, but fast per-core performance matters more as the world gets busier. Recruited tribesmen and large bases add work every tick.

Disk and bandwidth

Keep about 20GB free for the server, the live database and backups. Traffic is only about 100 kbps per player each way, so bandwidth is rarely the bottleneck.

Player slots

Slots only control how many people may join. They do not reserve RAM, fix CPU limits or stop an old world from outgrowing a cheap plan.

The official requirement and what it really means

The developer guidance for Soulmask dedicated servers is clear on the important part: one game process needs at least 16GB of memory and 2-4 CPU cores. One map equals one process, so Cloud Mist Forest and Shifting Sands are not two checkboxes inside the same lightweight server. Each map has its own launch script, its own memory footprint, its own ports and its own save data, on either Linux or Windows.

That official 16GB line is why most renters should treat 16GB as the safe choice, but it does not mean a 12GB plan is automatically impossible. Soulmask hosts put the server at about 11-12GB just to start, and usage climbs from there as the world develops. That is why a small private world can run on 12GB, and why most hosts tell busier servers to expect 16GB or more. The catch is headroom. If the server has almost no spare memory after startup, it has nowhere to put world growth, more NPC state, larger saves or the known memory leak, so the plan may look fine at launch and become unstable later.

Do not mix this up with the Steam store page for the game client. The store lists 16GB minimum and 24GB recommended for players’ PCs, plus 25GB storage, and those figures are for running the game on a local machine, not for hosting the dedicated server process.

The rest of the official numbers matter too: about 20GB free disk space, roughly 100 kbps per player in both directions, and two UDP ports per process. The game port is 8777 and the Steam query port is 27015. The telnet maintenance port 18888 is for local use and does not need to be opened publicly. If you are deciding between renting and using spare hardware at home, read our self-hosting explainer before you assume a spare desktop can carry an older persistent world without babysitting.

Here is what that looks like once the server has booted. The orange part is what the process needs before anyone has done much; the dark part is what is left for everything that happens afterwards.

What is left after the server boots
The full bar is the plan limit. Boot figure from Soulmask hosts’ own documentation.
12GB plan, just after boot
Boot: about 11-12GB
Almost nothing left for world growth, saves or memory creep.
12GB plan, days without a restart
Boot + world growth + memory creep
over the limit
The process runs into the limit: refused joins or a crash, often during busy hours.
16GB plan, just after boot
Boot: about 11-12GB
about 4GB spare
Room for tribesmen, bigger bases and save spikes.
16GB plan with a daily restart
Boot
growth + creep
spare
The restart clears the creep before it reaches the limit.
A 12GB plan can boot Soulmask, but 16GB is what gives a growing world somewhere to go. Growth and creep segments are drawn for illustration.

Also keep normal startup time in mind. Soulmask can take 2-5 minutes to start and appear in the browser, so a slow first boot is not automatically a crash.

What actually uses the resources

Soulmask is sized by world activity, not by a sales-page slot count. The server is continuously simulating persistent state, writing that state to disk, and sending updates to connected players. What matters is not only how many people are online, but how much the world contains and how much AI it has to think about every tick.

The load matrix below shows which world features mainly hit RAM, main-thread CPU, disk or bandwidth.

What each part of the world costs the server
What growsRAMCPU every tickDisk and savesBandwidth
Players onlineYesYesPartlyYes
Recruited tribesmenPartlyYesPartlyNo
Large basesYesYesPartlyNo
World ageYesPartlyYesNo
Uptime since restartYesNoNoNo
A second mapYesYesYesYes
Uptime: the known memory leak. A second map is a whole extra server process with its own allocation.
Bandwidth is about 100 kbps per player each way, so it rarely decides the plan.
Tribesmen and bases load the CPU even when few players are online, which is why slot counts are a poor guide to plan size.
  • Players: every online player adds movement, inventory state, combat state, crafting activity and network updates. That costs RAM for live state, CPU for simulation, and bandwidth for position and world updates. If usage rises mainly during peak hours and falls back when people leave, players are a real part of the load.
  • Recruited tribesmen: this is the big one people underestimate. Every recruited tribesman means more pathfinding, task logic and state checks every tick. That is mostly main-thread CPU, with extra RAM for persistent NPC data. If performance gets worse when people bring workers or guards into play, AI is usually the reason.
  • Bases and placed objects: large settlements mean more structures, containers and other placed objects for the server to keep track of, save and revisit. That raises RAM use and increases tick work around those areas. If lag is much worse near one base than everywhere else, the base itself is often the problem.
  • World age: an older world tends to carry more persistent state. More bases, more NPCs, more stored items and more accumulated database contents mean higher memory use at boot, longer saves and more chance that a plan which felt fine at the start will later feel cramped.
  • Saves and backups: writing the world database to disk costs I/O and can cause short hitches, especially on a busy persistent world. More frequent saves reduce rollback risk after a crash, but they create more frequent write bursts.
  • The memory leak: Soulmask servers are known to creep upward in RAM use over time, even with nobody online. This does not usually show up as immediate lag. It shows up as a server that was fine after a restart and later starts refusing joins or crashing near the plan limit.

The important consequence is simple: two servers with the same slot cap can need very different plans. A quiet low-build world with few tribesmen can be light, while a smaller group with aggressive recruitment and one huge settlement can be much heavier.

How much RAM a Soulmask server needs by group size

Buy for the world you expect in a month, not the empty map on day one. Player count is a starting point; how many tribesmen your group recruits and how big the settlements get decide where in each band you land.

Server type Players RAM CPU note What to watch
Small private tribe 2-4 12GB Usually workable if tribesmen and builds stay modest Memory after boot and memory creep between restarts
Active tribe or shared private world 4-8 16GB Better buffer once the world persists and AI grows Save hitches, peak-hour joins, rising usage over several days
Large persistent world 8+ 16GB plus headroom Fast cores matter as much as extra memory Old saves, recruited workforce, large settlements
Two-map cluster Varies Two separate allocations Each map is its own 2-4 core process Main server uptime, doubled memory footprint, doubled save load

The reason tribesmen can matter as much as players is that they keep the world busy even when the player count looks modest. Six people with a large recruited workforce can load the server harder than a bigger group with simple settlements and fewer active NPCs. If you want the broader logic behind headroom, ageing sandbox worlds and why dedicated servers rarely scale in a neat line with slot counts, our general hardware requirements explainer covers that background.

CPU: why fast cores beat more cores

The official guidance asks for 2-4 cores per process, and the reason it stops there is that a Soulmask server is not a pile of separate jobs that spread neatly across many cores. Much of the important work is the live world simulation: movement, combat, AI decision-making, interaction checks and structure state. That sort of work has a busy path every tick, so once the world gets active, the quality of the fastest core matters more than adding idle cores the process cannot use efficiently.

Extra cores still help. They give the operating system, networking, save tasks and other background work room to breathe, and they matter if the same machine is hosting several processes. But for one busy map, better per-core performance is what players tend to feel first. Large bases and recruited tribesmen both add more work to each tick, so CPU trouble usually shows up when people gather in developed areas or when many NPCs are active at once.

How CPU trouble looks different from RAM trouble

  • CPU-bound server: rubber-banding, delayed interactions and worse performance near specific settlements or during busy hours. The server often stays online, but the world feels sluggish.
  • RAM-bound server: steadily rising memory use, failed joins, sudden crashes or a process that becomes unreliable after long uptime even when nobody is doing much.

If a change really helped CPU pressure, you should notice fewer hitches around active tribesmen and big bases without much change to idle memory use. If memory was the real limit, CPU symptoms may stay similar but crashes and refused connections stop after adding headroom or restarting more sensibly.

Player slots do not size the server

Soulmask ships with a player cap of 20 in Engine.ini. That number only decides how many people can be online at once. It does not reserve any memory or CPU for them.

[Dedicated.Settings]
MaxPlayers=20

You can raise or lower that limit with -MaxPlayers, and Soulmask hosts document a ceiling of 70 players per server. That does not mean a busy 70-slot world will run on the same RAM as a tidy 20-slot private world. Slots only decide how many people may attempt to join; the real load still comes from active players, NPC crews, base sprawl, world age and uptime.

Bandwidth is the least confusing part once you do the arithmetic from the official figure. At about 100 kbps per player each way, 6 players use about 0.6 Mbps, 20 use about 2 Mbps, 40 use about 4 Mbps and 70 use about 7 Mbps each way. In practice that means bandwidth is rarely the thing that forces an upgrade. RAM and CPU are the usual constraints.

If you are comparing one plan sold by slots with another sold by memory, our slots vs RAM pricing explainer is the useful way to decode the sales pitch. For Soulmask, the safer buying rule is to choose enough RAM for the world you expect, then set the slot cap conservatively until you see how the world behaves under real use.

Disk space, saves and clean shutdowns

The server package itself is only about 1-2GB, but the official guidance asks for about 20GB of free disk space. That room goes on updates, the live world database and the backup copies that pile up over time. Disk is rarely the first bottleneck, but running short of space is how backups quietly stop working.

The main files worth knowing are WS/Saved/Worlds/Dedicated/Level01_Main/world.db for the world database and WS/Saved/Logs/WS.log for startup and error clues. Those paths matter because “it crashed” is vague; the log timing often tells you whether the fault happened during startup, save, backup or a period of heavy AI load.

The shipped defaults are saving=600 (game objects written to the database every 10 minutes) and backup=900 (the database written to disk every 15 minutes). A crash can therefore cost up to 10 minutes of play. Lowering the save interval to 300 means less progress can be lost in a crash, but it also means more frequent database writes, which can create more noticeable hitches on a busy old world. That trade-off is worth making on some private servers and not worth making on others; the point is to understand that you are exchanging smoother uptime for a shorter rollback window.

Backups also take disk space over time, so do not think only about the live database. Keep an eye on retained backup copies, and use a sensible retention routine rather than hoarding every file forever. If you want the broader housekeeping side of that, our game server backups guide covers the maintenance habits that stop save folders becoming a mess.

For shutdowns on your own machine, connect with telnet to 127.0.0.1 18888 and send quit 180. On a rented server, the panel’s Stop button does the same job. A clean shutdown gives the process time to finish writing the world; closing the window directly is how you end up with a rollback after the restart.

Two maps and clusters: treat them as two full servers

Soulmask has two maps, Level01_Main for Cloud Mist Forest and DLC_Level01_Main for Shifting Sands. One process hosts one map. If you want both maps available with character travel between them, you are building a cluster, not stretching one server a bit further.

That cluster needs a main server using -serverid and -mainserverport to manage player accounts, plus the client map servers that connect to it. The practical consequence is easy to miss: each map is still a full process with its own memory, CPU load, ports, save behaviour and restart schedule. If the main server is offline, the client servers stop working properly. Changing a server ID after players have joined can also cut them off from their characters.

So the budgeting rule is simple: both maps mean two full server allocations, not one plan with a small add-on. If your group only needs one map, keep one process and keep the budget simple.

Keeping a right-sized server stable

Schedule a daily restart for the memory leak

Soulmask has a known memory leak, so a scheduled daily restart in off-peak hours is normal maintenance, even on a generously sized plan. What the restart actually does is reset the process before leaked allocations can climb to the plan limit. The trade-off is obvious: a short daily interruption. The benefit is that you stop “fine yesterday, dead today” behaviour before it starts. You can tell it worked if memory drops back after each restart and stays safely below the limit until the next one.

Trim tribesmen limits before you buy a bigger plan

The highest-impact tuning usually lives in WS/Saved/GameplaySettings/GameXishu.json. These are the defaults:

{
  "GeRenMaxZhaoMuCount": 6,
  "GongHuiMaxZhaoMuCount": 50,
  "ManRenChuZhanCount": 3
}

Lowering those values works because it cuts the number of recruited NPCs the server has to keep alive in memory and process every tick for pathfinding and task logic. The trade-off for players is fewer workers, guards or travelling companions. If that change helps, you should see fewer peak-hour stalls and less area lag when multiple people have tribesmen active.

The other habit that matters is observation. Do not judge a plan only by the first evening after launch. Watch memory after boot, after a busy session and just before the next scheduled restart for several days. That tells you whether the plan is merely booting or actually carrying the world cleanly.

Situation Tune it first Upgrade when
Small private world on 12GB is stable straight after restart Keep the daily restart and watch memory growth over time Move to 16GB if usage drifts close to the limit before the next restart
Lag spikes happen mainly when people deploy tribesmen Lower the recruitment and deployment caps in GameXishu.json Upgrade CPU or move to a roomier plan if modest caps still leave the world sluggish
Only one huge base causes problems Trim placed objects and reduce complexity in that settlement Upgrade only if simple areas are also laggy and the issue is clearly world-wide
Everyone drops during saves or backup windows Check save timing, return to 600 if 300 is too aggressive, and review world bloat Upgrade if default save timing still causes severe disruption on an otherwise healthy server
The server crashes after long uptime but feels fine after a restart Schedule a daily restart and verify memory creep Add RAM if the process still runs into the limit before the planned restart window
You want both maps with travel between them Do not try to squeeze both into one process; build a proper cluster Upgrade the overall budget, because each map needs its own full allocation
Need a sensible Soulmask plan, not a slot number?

Once you size the server by world age, recruited tribesmen and whether you want one map or a cluster, plan shopping gets much simpler. Compare on memory, CPU room and restart options first, then use slots as a cap setting rather than a sizing tool.

Compare the best Soulmask server hosts

What 12GB+ plans cost

If you have decided you need 12GB or more, these are the regular monthly prices of the cheapest plans that start at that size, rather than plans you would have to upgrade straight away.

Cheapest Soulmask plans that start at 12GB or more
Regular monthly price of each host’s smallest plan at this size, November 2026
Sparked Host · 12GB
£9.02/mo ($12.00)
DatHost · 16GB
£12.45/mo ($16.56)
Host Havoc · 12GB
£18.80/mo ($25.00)
PingPlayers · 12GB
£22.92/30 days ($30.49)
Physgun · 14GB
£26.68/mo ($35.49)
Conversion used: £1 = $1.33
DatHost is the cheapest plan here that meets the official 16GB figure out of the box.

See the full Soulmask host comparison with all 40 hosts, UK locations, backups and refund terms.

Is it RAM, CPU or something else? Troubleshooting by symptom

Use the flow below to separate RAM pressure from CPU pressure before you start changing settings at random.

Is it RAM, CPU or something else?
Your Soulmask server lags, crashes or refuses joins
Is memory near the plan limit, or still rising with nobody online?
Yes
RAMRestart and schedule a daily restart. Near the limit again soon after boot: move up to 16GB.
No
Is lag worst when tribesmen are active, or near one big base?
Yes
CPU every tickLower the tribesmen limits in GameXishu.json or trim that base. Still sluggish: a faster CPU.
No
Does everyone drop or freeze at the same moment?
Yes
Saves or memoryLine the drops up with save times in WS.log. If you lowered saving to 300, go back to 600.
No
Is the server missing from the browser after 5 minutes?
Yes
PortsCheck that UDP 8777 and 27015 are open and forwarded.
No
Probably one player’s connection or PC. Check with another player before changing the server.
Stop at the first yes. Most Soulmask stability problems are memory, and most lag problems are tribesmen or one oversized base.
Symptom Likely cause First fix
Server will not start on a tiny plan Not enough free RAM even for boot Move to a plan with at least 12GB; the server needs about 11-12GB just to start
Server starts, then refuses new connections later Memory has crept up near the plan limit Restart the process and add headroom if the pattern repeats
Memory keeps rising even when nobody is online Known memory leak Schedule a daily restart in off-peak hours
Everyone disconnects at the same moment Auto-save hitch on a large world, or low-memory instability Check save timing and memory use around the event
Lag is much worse near one settlement That base has too many placed objects Trim the build before buying more RAM
Peak-time stutter appears when several players have tribesmen active AI pathfinding and task logic are saturating CPU every tick Lower tribesmen caps in GameXishu.json
Server does not appear in the browser Ports 8777 or 27015 are not reachable Check that both UDP ports are open and correctly forwarded
Players lose progress after a restart Dirty shutdown or crash before the next save Use quit 180 for shutdowns and review the save interval
Characters cannot access the other map in a cluster Main server is offline, or the server ID changed after players joined Restore the main server and keep the existing server ID
The server takes several minutes to show up after start Normal startup delay Wait through the usual 2-5 minute window and watch WS.log

If a symptom still points both ways, line up memory growth with save timing in WS.log. A crash after steady RAM creep is a different problem from a repeatable hitch near one oversized base.

Soulmask server requirements FAQ

How much RAM does a Soulmask server need?

The official guidance is 16GB per game process, with one map equalling one process. In practice, a small fresh world can boot on about 11-12GB, which is why 12GB plans can work for private groups. The reason 16GB is the safer target is headroom for world growth, recruited tribesmen, saves and the known memory leak.

Is 8GB enough for a Soulmask server?

Not for a world you plan to keep. Soulmask hosts put the server at about 11-12GB just to start, so an 8GB plan has no headroom and tends to crash or refuse joins as soon as the world develops. Treat 12GB as the realistic starting point for a small group.

How many players can a Soulmask server hold?

The shipped default is MaxPlayers=20, and you can change that with -MaxPlayers. Hosts document a ceiling of 70 players per server, but that is only a slot ceiling, not a guarantee of smooth performance. Practical capacity depends much more on tribesmen AI, base size, world age and the CPU available to the process.

Why does my Soulmask server use more RAM the longer it runs?

Part of that is ordinary world growth, because older worlds hold more persistent state. Part of it is a known memory leak, which makes usage creep upward even with nobody online. That is why a daily scheduled restart is recommended even on servers that otherwise seem healthy.

Do I need a second server to run both Soulmask maps?

Yes, if you want both maps available properly. Each map is its own server process with its own RAM, CPU, ports and save files, and a cluster also needs a main server for account handling. In budgeting terms, both maps mean two full allocations, not one server with a light add-on.

More Soulmask and server sizing guides

Compare Soulmask hosts and dig into the hardware basics behind this guide.

Game updates, defaults and host panels change. Where your host’s current panel works differently from what is described here, follow the panel.